@PaulMSherman Not every justice gets an opportunity to leave their mark on the popular view of the court like a BREAD Political Action Committee v. FEC
@tracewoodgrains Griggs is a good case for the (understandably) oft-derided Anthony Kennedy school of making up a fake rule to get to the right outcome and then spending 15 pages making it clear this only applies to this one case
@tracewoodgrains The hardest part about this for me at my (likely even further left law school) was just the absence of any is-aught decoupling, respect for formalism, etc. Maybe the very essence of the law in a liberal society! Just totally subsumed by conflict theory.
@TheZvi Feels like a little more evidence to me that the continual learning path anthropic has been vagueposting about this year is just going to be better and bigger versions of current in context learning paradigm
@RhythmDoc01@damintoell Even if the court had the power to disqualify a pardon on these grounds (it doesn’t), no Supreme Court in history (least of all this court) would ever try to use it!
@RatOrthodox But maybe those objections go away if the vision is just to create a tool that helps the slice of weird rat-ish types who would ever want to use this sort out narrow object level disputes
@RatOrthodox The main reason I assume this won’t work, which I was surprised to not really find in Scott’s piece, is some combo of “getting two people who disagree to use an app like this in good faith is 95% of the battle”, and “interest in this app self selects for a ton of commonalities”
@peterwildeford For better and also for worse, Sanders continues to be the ~only politician in Washington that you can model as “going about the world applying his beliefs and observations in good faith” rather than doing some other thing
@tenobrus@theo Maybe the perspective is that Anth doesn’t respect/prioritize engineers, ergo they’ve proven themselves evil and not to be trusted with superintelligence, which feels to me like an anachronistic port from the Apple hates hackers, go to Linux/open source/modular hardware days
@tenobrus@theo Continually confused by the amount of overlap between scaling will hold/asi soon/straight lines on a graph and ‘ai as normal technology’ beliefs. Zuck seems to think approx. “this is the most important technology ever, AGI might someday improve the reels algo. by 7, even 8%”
@tenobrus Maybe I’m just not bitter lesson pilled enough but I’m starting to think having a vaguely talkie-shaped model as a rewrite layer might be the only way out of sloppocolypse for at least the next year or two
@ASFleischman Is the lack of deference here mostly down to the fact that actual malice is a first-amendment derived standard and thereby not just normal fact-finding? Or would a criminal jury still get more deference in, e.g., an obscenity/CSAM trial or some 4th am. issue or something?